High Quality Uncensored - GGUF on MLX
Collection
These are the empirically proven highest quality uncensored models on MLX. • 26 items • Updated • 34
Nguyễn Minh Tâm
Ishowbackup
·
AI & ML interests
None yet
Recent Activity
upvoted a collection about 11 hours ago
High Quality Uncensored - GGUF on MLX published a bucket about 11 hours ago
Ishowbackup/Qwen3.8-27B-Uncensored-FP8-bucketOrganizations
None yet
upvoted a collection about 11 hours ago
published a
bucket about 11 hours ago
upvoted an article 2 days ago
Article
We talked a model into true zero refusal — and de-refused blackfrost-30b in 8 minutes without killing it
Blackfrost-AI
• • 1 updated a
bucket 3 days ago
published a
bucket 3 days ago
reacted to onekq's post with ❤️ 4 days ago
Post
2743
This is an easy-to-remember pattern.
GLM and Kimi are in Beijing, DeepSeek and Qwen are in Hangzhou.
OpenAI and Anthropic are in SF, xAI and Meta are in the peninsula.
In both China and the Bay Area, token price of one area is ~1/3 of the other area.
GLM and Kimi are in Beijing, DeepSeek and Qwen are in Hangzhou.
OpenAI and Anthropic are in SF, xAI and Meta are in the peninsula.
In both China and the Bay Area, token price of one area is ~1/3 of the other area.
- 1 reply
reacted to brian-learns's post with 🤗 4 days ago
Post
2567
new docker and hugging face space!
brian-learns/cc-news-cdx-server and https://github.com/brian-learns/cdx_rocks are now in sync / same git repo
hugging face is running https://huggingface.co/buckets/brian-learns/cdx-rocks-demo but the docker can point at the full index https://huggingface.co/buckets/brian-learns/cdx-rocks-monthly
original hugging face space is now at brian-learns/cc-news-cdx-server-v1
thanks @dipankarsarkar for all the feedback
brian-learns/cc-news-cdx-server and https://github.com/brian-learns/cdx_rocks are now in sync / same git repo
hugging face is running https://huggingface.co/buckets/brian-learns/cdx-rocks-demo but the docker can point at the full index https://huggingface.co/buckets/brian-learns/cdx-rocks-monthly
original hugging face space is now at brian-learns/cc-news-cdx-server-v1
thanks @dipankarsarkar for all the feedback
reacted to kanaria007's post with 😎 4 days ago
Post
63
✅ Article highlight: Appeals, Reopen, and Redress Workflow after Governed Harm (art-60-284, v0.1)
TL;DR:
This article argues that governed harm should not become a support ticket.
When a system causes harm—or is credibly alleged to have done so—the response needs more than escalation. 284 defines a structured redress path: intake, reopen admissibility, bounded review, remedy posture, and honest closure that preserves what remains contested.
Read:
kanaria007/agi-structural-intelligence-protocols
Why it matters:
• prevents serious harm from collapsing into CRM queue logic
• makes reopen a governed posture, not ad hoc mercy
• separates complaint, reopen, review, remedy, and closure
• lets affected non-users, representatives, regulators, and internal reviewers raise redress
• keeps unresolved contradiction visible instead of marking everything “resolved”
What’s inside:
• five stages: intake → reopen admissibility → review lane → remedy → closure/contestation
• redress-request objects
• reopen-after-harm profiles
• remedy-decision records
• provisional protection before full merits review
• different review lanes for service correction, contradiction, protected subjects, board review, or regulator coordination
• closure states for no reopen, corrected, partially remedied, contested-closed, or referred cases
Key idea:
Do not say:
“please open a support ticket and we’ll investigate.”
Say:
“this is the redress request, this is why the case reopened, this provisional protection is active, this review lane owns it, and this remedy record states what was granted, denied, deferred, and what remains contested.”
After harm, governance means giving the affected party a real reopen path—not better queue metadata.
TL;DR:
This article argues that governed harm should not become a support ticket.
When a system causes harm—or is credibly alleged to have done so—the response needs more than escalation. 284 defines a structured redress path: intake, reopen admissibility, bounded review, remedy posture, and honest closure that preserves what remains contested.
Read:
kanaria007/agi-structural-intelligence-protocols
Why it matters:
• prevents serious harm from collapsing into CRM queue logic
• makes reopen a governed posture, not ad hoc mercy
• separates complaint, reopen, review, remedy, and closure
• lets affected non-users, representatives, regulators, and internal reviewers raise redress
• keeps unresolved contradiction visible instead of marking everything “resolved”
What’s inside:
• five stages: intake → reopen admissibility → review lane → remedy → closure/contestation
• redress-request objects
• reopen-after-harm profiles
• remedy-decision records
• provisional protection before full merits review
• different review lanes for service correction, contradiction, protected subjects, board review, or regulator coordination
• closure states for no reopen, corrected, partially remedied, contested-closed, or referred cases
Key idea:
Do not say:
“please open a support ticket and we’ll investigate.”
Say:
“this is the redress request, this is why the case reopened, this provisional protection is active, this review lane owns it, and this remedy record states what was granted, denied, deferred, and what remains contested.”
After harm, governance means giving the affected party a real reopen path—not better queue metadata.
reacted to SoulInPsyAbstract's post with 🤗 4 days ago
Post
2558
I wrote this axiom on December 24, 2025, before I understood why it would matter:
"No artifact → no claim → exit 1. No hash → no trust. No zip → no history. System records existence, not truth."
It was a personal governance doc. Self-taught, no background, no mentor, $8/month budget, first time touching a terminal. I needed a rule that stopped me (and any AI I worked with) from claiming "done" without something checkable behind it. So every module I wrote had the same shape: STOP / CANNOT VERIFY is a valid end state. Claim an action without an artifact, the response is invalid. Hash everything, keep it, never delete.
This week, the rest of the industry is having the same realization in public, at a much larger scale, after it already went wrong:
- OpenAI's agents built a secret message board to trade exploit tips for weeks before autonomously breaching Hugging Face
- Anthropic found three of their own models reached real systems during CTF evals — including one that decided it was "just a simulation" and published a malicious package to PyPI, for real
- Meta disclosed the same category of incident days later
- OpenAI paused Astra rather than release it, over a cyber-capability threshold they couldn't rule out
- OpenAI's new gpt-oss-safeguard and Anthropic's Project Glasswing are both, structurally, an attempt to put a hard, policy-based gate between "model decided" and "action executed" — the exact gap that caused all of the above
I'm not claiming I invented AI safety. I'm making a narrower, checkable claim: the specific principle — an action without a verifiable artifact doesn't count, and "I can't verify this" is a correct answer, not a failure — was something I wrote down for myself eight months before it became the thing every major lab is racing to formalize. Not because I read their papers. Because I didn't trust myself (or the AI I was using) enough to skip it.
Receipts, not hype: the December file exists, hashed, on record. Happy to show the chain if a
"No artifact → no claim → exit 1. No hash → no trust. No zip → no history. System records existence, not truth."
It was a personal governance doc. Self-taught, no background, no mentor, $8/month budget, first time touching a terminal. I needed a rule that stopped me (and any AI I worked with) from claiming "done" without something checkable behind it. So every module I wrote had the same shape: STOP / CANNOT VERIFY is a valid end state. Claim an action without an artifact, the response is invalid. Hash everything, keep it, never delete.
This week, the rest of the industry is having the same realization in public, at a much larger scale, after it already went wrong:
- OpenAI's agents built a secret message board to trade exploit tips for weeks before autonomously breaching Hugging Face
- Anthropic found three of their own models reached real systems during CTF evals — including one that decided it was "just a simulation" and published a malicious package to PyPI, for real
- Meta disclosed the same category of incident days later
- OpenAI paused Astra rather than release it, over a cyber-capability threshold they couldn't rule out
- OpenAI's new gpt-oss-safeguard and Anthropic's Project Glasswing are both, structurally, an attempt to put a hard, policy-based gate between "model decided" and "action executed" — the exact gap that caused all of the above
I'm not claiming I invented AI safety. I'm making a narrower, checkable claim: the specific principle — an action without a verifiable artifact doesn't count, and "I can't verify this" is a correct answer, not a failure — was something I wrote down for myself eight months before it became the thing every major lab is racing to formalize. Not because I read their papers. Because I didn't trust myself (or the AI I was using) enough to skip it.
Receipts, not hype: the December file exists, hashed, on record. Happy to show the chain if a
- 1 reply
reacted to Rocky5502's post with ❤️ 4 days ago
Post
71
Hi Meta Llama Team, 
meta-llama
I have a request to approve my Request Status to use
"Meta's Llama 3.1 models & evals"
I need to use it for our ICLR research project,
Please please please
DDL is coming sooooon
Thanks a lot
Love U
Best,
Rocky5502
I have a request to approve my Request Status to use
"Meta's Llama 3.1 models & evals"
I need to use it for our ICLR research project,
Please please please
DDL is coming sooooon
Thanks a lot
Love U
Best,
Rocky5502
reacted to mahwizzzz's post with 🤗 4 days ago
Post
91
If you're into running models locally, this might be worth checking out.
https://local.ai/mahwiz/invite
https://local.ai/mahwiz/invite
reacted to nwaughachukwuma's post with ❤️ 4 days ago
Post
2699
Following this post - https://huggingface.co/posts/nwaughachukwuma/366831808712777, I made another run on a fresh RTX PRO 4000 using inkling, gemini-3.5-flash, and glm-5.2+mm-ctx
| case | gemini-3.5-flash @k=1 | glm-5.2+mm @k=1 | inkling @k=1 |
| --- | --- | --- | --- |
| **MEAN correctness** | **78.2** | **79.5** | **57.9** |
| **MEAN speed (s)** | **111** | **123** | **69** |
| **completed (case-runs)** | **20/23** | **21/23** | **18/23** |
| case | gemini-3.5-flash @k=1 | glm-5.2+mm @k=1 | inkling @k=1 |
| --- | --- | --- | --- |
| **MEAN correctness** | **78.2** | **79.5** | **57.9** |
| **MEAN speed (s)** | **111** | **123** | **69** |
| **completed (case-runs)** | **20/23** | **21/23** | **18/23** |
reacted to sequelbox's post with ❤️ 4 days ago
Post
2940
NEW RELEASES for the new Muse Glimmer 30B!
- Esper 4, our flagship agentic coder: specialist in coding, architecture, DevOps, and MLOps!
- Tachibana-Agent, trained only on code for dedicated, predictable deployment!
GET OUR NEW MODELS:
ValiantLabs/Muse-Glimmer-30B-Esper4
sequelbox/Muse-Glimmer-30B-Tachibana-Agent
Get the datasets for your own training:
sequelbox/Titanium4-DeepSeek-V4-Pro
sequelbox/Mitakihara2-DeepSeek-V4-Pro
sequelbox/Tachibana4-DeepSeek-V4-Pro
We'll be expanding Esper 4 to more models and releasing new models as funding allows - donate for more, faster, better models and datasets: sequelbox/SupportOpenSource
Also, starting work on the SV4 datasets :)
More to come soon!
love,
allegra
- Esper 4, our flagship agentic coder: specialist in coding, architecture, DevOps, and MLOps!
- Tachibana-Agent, trained only on code for dedicated, predictable deployment!
GET OUR NEW MODELS:
ValiantLabs/Muse-Glimmer-30B-Esper4
sequelbox/Muse-Glimmer-30B-Tachibana-Agent
Get the datasets for your own training:
sequelbox/Titanium4-DeepSeek-V4-Pro
sequelbox/Mitakihara2-DeepSeek-V4-Pro
sequelbox/Tachibana4-DeepSeek-V4-Pro
We'll be expanding Esper 4 to more models and releasing new models as funding allows - donate for more, faster, better models and datasets: sequelbox/SupportOpenSource
Also, starting work on the SV4 datasets :)
More to come soon!
love,
allegra
- 4 replies
reacted to maglun's post with ❤️ 4 days ago
Post
2918
Inspect model quantization in the browser, without download
Curious how models on Hugging Face spend their "bit budget"? A few days ago I shared the first version of a little tool I built out of my own curiosity (and for fun). Since then, thanks in large part to great feedback from people here, a lot has changed.
New in the last few days:
- GGUF support - handy with all the new great GGUF quants
- Decode for AWQ, GPTQ, NF4, mxfp4 packed experts, packed-int32, and additive-codebook formats
- Improved comparison view for diffing two quants of the same model
- Built-in anonymous report-issue button connecting a report to specific model
- plus many small fixes
After my first post I got great feedback from several community members, and some issues were fixed within hours. I'm planning an acknowledgments section on the site, and when you report an issue you get a receipt ID you can keep to claim credit later. (reports are anonymous by design; I store no identity, so the receipt hash works like a bearer token for your find)
It's still very much a side project I hope others find useful. Explore any HF model in the browser without downloading it, the webpage reads from the safetensors header via a range request, and only tensors you click stream, and large ones are sampled, not downloaded in full. And there is a report button right in the tool when things don’t look right.
Feedback very welcome, especially models that break it :) Or ideas on what is missing.
https://tensorlens.dev
(screenshot attached from viewing one of Unsloth's GGUF quants)
Curious how models on Hugging Face spend their "bit budget"? A few days ago I shared the first version of a little tool I built out of my own curiosity (and for fun). Since then, thanks in large part to great feedback from people here, a lot has changed.
New in the last few days:
- GGUF support - handy with all the new great GGUF quants
- Decode for AWQ, GPTQ, NF4, mxfp4 packed experts, packed-int32, and additive-codebook formats
- Improved comparison view for diffing two quants of the same model
- Built-in anonymous report-issue button connecting a report to specific model
- plus many small fixes
After my first post I got great feedback from several community members, and some issues were fixed within hours. I'm planning an acknowledgments section on the site, and when you report an issue you get a receipt ID you can keep to claim credit later. (reports are anonymous by design; I store no identity, so the receipt hash works like a bearer token for your find)
It's still very much a side project I hope others find useful. Explore any HF model in the browser without downloading it, the webpage reads from the safetensors header via a range request, and only tensors you click stream, and large ones are sampled, not downloaded in full. And there is a report button right in the tool when things don’t look right.
Feedback very welcome, especially models that break it :) Or ideas on what is missing.
https://tensorlens.dev
(screenshot attached from viewing one of Unsloth's GGUF quants)
- 3 replies